Why the Common Ad Metrics Don’t Work Anymore and What We Did About It

A custom metric helped identify purchase intent for Ford customers

Industry voices have long argued that digital marketing activity should have a clear and measurable connection to business outcomes. So why do the majority of marketers still rely on outdated proxy metrics such as cost-per-click (CPC), cost-per-action (CPA), click-through-rate (CTR), conversion rates (CVR), or obscure viewability benchmarks?

Now more than ever, metrics that bear weak correlations to real business results can yield misleading data and result in misguided tactics. For example, by making ad formats more intrusive and harder for consumers to avoid, advertisers can gain more clicks. Ask yourself, are those ‘accidental’ clicks likely to deliver more sales?

On the other hand, higher-level marketing metrics – for instance brand impact, customer retention, and lifetime value – can be closely linked to business success. These aren’t typically applied to lower funnel digital strategies to drive sales, but we can still pursue the same concept: chasing metrics that directly impact the business instead of proxies that are simpler and more immediate.

Programmatic media, when combined with ample data and Artificial Intelligence (AI) technology, can deliver greater insight into business success. We can observe, analyze, and conclude decisively what works for each client based on how its audiences act and which business goals it wants to achieve. We can integrate those learnings into each ad campaign, each media buy, and each bid and engineer outcomes that are intrinsically valuable to the client.

That’s a far cry from blasting more impressions, gaming more clicks, and hoping for a correlation to sales. And most importantly, it acknowledges the individual needs and circumstances of every client.

Using these insights, we explored a custom approach for the automotive manufacturer Ford and its agency, GTB, in the Netherlands and Norway.

Together we analyzed customer behaviors on the Ford website and developed a statistical model that accurately quantified the value of key website actions – such as booking a test drive, downloading a brochure, and finding the nearest dealership – and linked them to their impact on offline car sales. We built an algorithm that evaluated each visit and action on the Ford site and determined its likelihood to deliver a new car order. A bespoke metric was created, Purchase Intent (PI), and a PI score was assigned to each visitor on the Ford website to quantify their potential value as a likely buyer. As visitors took certain actions that indicated higher propensities to buy – such as requesting a test drive – the algorithm assigned them higher PI values. Visitors with higher PI scores warranted higher media investments in retargeting ad efforts.

Our AI technology, Copilot, was then used to drive the implementation of this PI metric. Using the customized optimization algorithms that accurately predicted any user’s PI score, Copilot allocated exactly the right media investment through programmatic bidding, in line with the potential PI value. Now Ford has a data-driven, highly scalable, statistically validated method to optimize media investments toward real car sales. And the effectiveness of programmatic activity only increases as PI is scaled across all EMEA markets.

Ford can now calculate the actual value of every euro it spends on media – not only in clicks or impressions, but in actual business results. That alone is a huge benefit, empowering greater clarity and confidence on future investments. Using PI Ford can drive a higher return on investment from its existing media budgets, and has achieved a 45% reduction in cost per conversions. As the use of PI continues and expands across the entire region, Ford will be able to further increase efficiency and improve performance, gaining a significant competitive edge.

Adam Swain, Biddable Partner at GTB, highlighted this,

Being able to marry Copilot with a simple, quantifiable metric linked to business success has been very valuable. It allows us to really focus on what works and optimize strategy accordingly. No more arguing over what metric to use!”

 

The potential of customized metrics is not limited to the automotive sector. Using the strategy of connecting “simple” conversion data with goals that are intrinsically valuable for a client, custom metrics will beat traditional measurements like CTR and CPA every time. For example, within events, real-time seat availability could be integrated into targeting strategies; for home security, governmental crime rate statistics based on location would help to reach more relevant consumers. The possibilities are endless.

The question then remains, why are advertisers still married to outdated proxy metrics? It’s important to caveat that customized metrics aren’t a one-size-fits-all solution. PI was a complex and substantial effort that required the expertise of multiple data science teams, the creation of custom AI algorithms, a willingness to invest in test-and-learn development, and a whole lot more.

PI is an aspirational example of outcome-driven media at its best. It demonstrates what’s possible when the ambition to advance advertising maturity is granted equal footing with the demand for immediate validation; when brands are willing to shed faulty preconceptions and prioritize meaningful results. This is something we should all strive for.

Industry voices have long argued that digital marketing activity should have a clear and measurable connection to business outcomes. So why do the majority of marketers still rely on outdated proxy metrics such as cost-per-click (CPC), cost-per-action (CPA), click-through-rate (CTR), conversion rates (CVR), or obscure viewability benchmarks?

Now more than ever, metrics that bear weak correlations to real business results can yield misleading data and result in misguided tactics. For example, by making ad formats more intrusive and harder for consumers to avoid, advertisers can gain more clicks. Ask yourself, are those ‘accidental’ clicks likely to deliver more sales?

On the other hand, higher-level marketing metrics – for instance brand impact, customer retention, and lifetime value – can be closely linked to business success. These aren’t typically applied to lower funnel digital strategies to drive sales, but we can still pursue the same concept: chasing metrics that directly impact the business instead of proxies that are simpler and more immediate.

Programmatic media, when combined with ample data and Artificial Intelligence (AI) technology, can deliver greater insight into business success. We can observe, analyze, and conclude decisively what works for each client based on how its audiences act and which business goals it wants to achieve. We can integrate those learnings into each ad campaign, each media buy, and each bid and engineer outcomes that are intrinsically valuable to the client.

That’s a far cry from blasting more impressions, gaming more clicks, and hoping for a correlation to sales. And most importantly, it acknowledges the individual needs and circumstances of every client.

Using these insights, we explored a custom approach for the automotive manufacturer Ford and its agency, GTB, in the Netherlands and Norway.

Together we analyzed customer behaviors on the Ford website and developed a statistical model that accurately quantified the value of key website actions – such as booking a test drive, downloading a brochure, and finding the nearest dealership – and linked them to their impact on offline car sales. We built an algorithm that evaluated each visit and action on the Ford site and determined its likelihood to deliver a new car order. A bespoke metric was created, Purchase Intent (PI), and a PI score was assigned to each visitor on the Ford website to quantify their potential value as a likely buyer. As visitors took certain actions that indicated higher propensities to buy – such as requesting a test drive – the algorithm assigned them higher PI values. Visitors with higher PI scores warranted higher media investments in retargeting ad efforts.

Our AI technology, Copilot, was then used to drive the implementation of this PI metric. Using the customized optimization algorithms that accurately predicted any user’s PI score, Copilot allocated exactly the right media investment through programmatic bidding, in line with the potential PI value. Now Ford has a data-driven, highly scalable, statistically validated method to optimize media investments toward real car sales. And the effectiveness of programmatic activity only increases as PI is scaled across all EMEA markets.

Ford can now calculate the actual value of every euro it spends on media – not only in clicks or impressions, but in actual business results. That alone is a huge benefit, empowering greater clarity and confidence on future investments. Using PI Ford can drive a higher return on investment from its existing media budgets, and has achieved a 45% reduction in cost per conversions. As the use of PI continues and expands across the entire region, Ford will be able to further increase efficiency and improve performance, gaining a significant competitive edge.

Adam Swain, Biddable Partner at GTB, highlighted this, “Being able to marry Copilot with a simple, quantifiable metric linked to business success has been very valuable. It allows us to really focus on what works and optimize strategy accordingly. No more arguing over what metric to use!”

The potential of customized metrics is not limited to the automotive sector. Using the strategy of connecting “simple” conversion data with goals that are intrinsically valuable for a client, custom metrics will beat traditional measurements like CTR and CPA every time. For example, within events, real-time seat availability could be integrated into targeting strategies; for home security, governmental crime rate statistics based on location would help to reach more relevant consumers. The possibilities are endless.

The question then remains, why are advertisers still married to outdated proxy metrics? It’s important to caveat that customized metrics aren’t a one-size-fits-all solution. PI was a complex and substantial effort that required the expertise of multiple data science teams, the creation of custom AI algorithms, a willingness to invest in test-and-learn development, and a whole lot more.

PI is an aspirational example of outcome-driven media at its best. It demonstrates what’s possible when the ambition to advance advertising maturity is granted equal footing with the demand for immediate validation; when brands are willing to shed faulty preconceptions and prioritize meaningful results. This is something we should all strive for.

Boris-headshot

Boris Olij, Global Accounts Strategist

Boris Olij is an award-winning digital marketeer who’s continuously looking to find new, effective, and fun ways to handle data-led marketing and advertising challenges. One of his goals is to simplify effective application of Reinforced Machine Learning, Artificial Intelligence and Robotic Process Automation within the programmatic advertising space – unpacking black-box algorithms to explain outcomes and solutions

James Simpson_bw

James Simpson, EMEA Client Lead

In his role of EMEA Client Lead at Xaxis, James is responsible for the overall growth, service and delivery of Ford across the region. Ford has become one of the most forward-thinking clients always at the forefront in testing programmatic first ideas, which has resulted in several award wins for custom AI solutions. James’s passion for programmatic advertising started at iProspect where he was able to gain experience in AI as well as cross channel performance experience. As an EMEA Client Lead, James key in building the relationship between Xaxis and client and is always pushing boundaries in the programmatic landscape.

Related Posts